When having a conversation about AI, there should be an equivalent to Godwin’s law – the principle that the longer an online discussion continues the higher the probability someone will make a comparison to Hitler. For you cannot hold a discussion about AI before someone mentions The Terminator, it’s just a matter of time.
While the idea we’re heading toward an inevitable conflict between man versus machine has taken root in the collective consciousness, there are actually many more lower-level concerns marketers should be aware of if they’re thinking of adding an AI solution to their tech stack.
“What the vendors will tell you – the single best thing about AI is you don’t even need to know how it works. It’ll just magically appear, you’ll press some buttons, and it’ll solve all your problems like magic,” Nick Worth, CMO at Selligent, told the audience at B2B InTech last month (pictured). “But the problem is it’s a lot harder than that. As you’re leaving these executional burdens behind, you still have to be very mindful of what the AI is going to do.”
He illustrates with the example of white socks. People buy a lot of white socks. Almost all of us wear socks, and a lot of them tend to be white. When products get recommended to consumers, the algorithm suggests white socks because they’re very popular. This isn’t particularly damaging, but it shows “you, the marketer, and the AI need to be a lot smarter about it, in thinking about what AI’s going to do for you,” says Nick.
More worryingly: “AI will racially profile customers because it doesn’t have morality, it has no ethical concerns,” he adds. “It doesn’t racially profile people because it doesn’t like people of a particular ethnicity or another, it’s just following the math. You have to be really careful it doesn’t do that. And furthermore, people don’t find out your brand has been racially profiling them that was never your intention in what you set up. You need to be super careful.”
"It’s a long road to that magical nirvana where you’ve got your feet on the desk with a long, cool drink"
Consider the human element
But the biggest consideration if you’re dropping a piece of AI into your stack isn’t the machine at all, but the humans. “You’re not going to have a fully realised AI experience – it is not turning autopilot on in a plane. You’re going to have lots of other human-driven components and it’s getting that integration right. Robots work better with human supervision than if left to their own devices. You have to stay on top of this, it doesn’t give you a free pass to let AI run everything for you,” Nick says.
“There’s enormous potential in machines doing all this work for us that we cannot do ourselves – in particular, the power of bringing together all these diverse data sets and targeting on a one-to-one level. But it’s a long road to that magical nirvana where you’ve got your feet on the desk with a long cool drink. There are many steps and lots of care that needs to be taken.”
